/*
* Licensed to Elasticsearch under one or more contributor
* license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright
* ownership. Elasticsearch licenses this file to you under
* the Apache License, Version 2.0 (the "License"); you may
* not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
package org.elasticsearch.cluster.routing.allocation;
import org.elasticsearch.Version;
import org.elasticsearch.cluster.ClusterState;
import org.elasticsearch.cluster.metadata.IndexMetaData;
import org.elasticsearch.cluster.metadata.MetaData;
import org.elasticsearch.cluster.node.DiscoveryNodes;
import org.elasticsearch.cluster.routing.RoutingTable;
import org.elasticsearch.cluster.routing.allocation.command.AllocationCommands;
import org.elasticsearch.cluster.routing.allocation.command.MoveAllocationCommand;
import org.elasticsearch.cluster.routing.allocation.decider.ClusterRebalanceAllocationDecider;
import org.elasticsearch.common.logging.ESLogger;
import org.elasticsearch.common.logging.Loggers;
import org.elasticsearch.test.ElasticsearchAllocationTestCase;
import org.junit.Test;
import static org.elasticsearch.cluster.routing.ShardRoutingState.*;
import static org.elasticsearch.common.settings.ImmutableSettings.settingsBuilder;
import static org.hamcrest.Matchers.equalTo;
/**
*/
public class DeadNodesAllocationTests extends ElasticsearchAllocationTestCase {
private final ESLogger logger = Loggers.getLogger(DeadNodesAllocationTests.class);
@Test
public void simpleDeadNodeOnStartedPrimaryShard() {
AllocationService allocation = createAllocationService(settingsBuilder()
.put("cluster.routing.allocation.concurrent_recoveries", 10)
.put(ClusterRebalanceAllocationDecider.CLUSTER_ROUTING_ALLOCATION_ALLOW_REBALANCE, "always")
.build());
logger.info("--> building initial routing table");
MetaData metaData = MetaData.builder()
.put(IndexMetaData.builder("test").settings(settings(Version.CURRENT)).numberOfShards(1).numberOfReplicas(1))
.build();
RoutingTable routingTable = RoutingTable.builder()
.addAsNew(metaData.index("test"))
.build();
ClusterState clusterState = ClusterState.builder(org.elasticsearch.cluster.ClusterName.DEFAULT).metaData(metaData).routingTable(routingTable).build();
logger.info("--> adding 2 nodes on same rack and do rerouting");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder()
.put(newNode("node1"))
.put(newNode("node2"))
).build();
RoutingAllocation.Result rerouteResult = allocation.reroute(clusterState);
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
// starting primaries
rerouteResult = allocation.applyStartedShards(clusterState, clusterState.routingNodes().shardsWithState(INITIALIZING));
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
// starting replicas
rerouteResult = allocation.applyStartedShards(clusterState, clusterState.routingNodes().shardsWithState(INITIALIZING));
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
logger.info("--> verifying all is allocated");
assertThat(clusterState.routingNodes().node("node1").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node1").get(0).state(), equalTo(STARTED));
assertThat(clusterState.routingNodes().node("node2").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node2").get(0).state(), equalTo(STARTED));
logger.info("--> fail node with primary");
String nodeIdToFail = clusterState.routingTable().index("test").shard(0).primaryShard().currentNodeId();
String nodeIdRemaining = nodeIdToFail.equals("node1") ? "node2" : "node1";
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder()
.put(newNode(nodeIdRemaining))
).build();
rerouteResult = allocation.reroute(clusterState);
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
assertThat(clusterState.routingNodes().node(nodeIdRemaining).get(0).primary(), equalTo(true));
assertThat(clusterState.routingNodes().node(nodeIdRemaining).get(0).state(), equalTo(STARTED));
}
@Test
public void deadNodeWhileRelocatingOnToNode() {
AllocationService allocation = createAllocationService(settingsBuilder()
.put("cluster.routing.allocation.concurrent_recoveries", 10)
.put(ClusterRebalanceAllocationDecider.CLUSTER_ROUTING_ALLOCATION_ALLOW_REBALANCE, "always")
.build());
logger.info("--> building initial routing table");
MetaData metaData = MetaData.builder()
.put(IndexMetaData.builder("test").settings(settings(Version.CURRENT)).numberOfShards(1).numberOfReplicas(1))
.build();
RoutingTable routingTable = RoutingTable.builder()
.addAsNew(metaData.index("test"))
.build();
ClusterState clusterState = ClusterState.builder(org.elasticsearch.cluster.ClusterName.DEFAULT).metaData(metaData).routingTable(routingTable).build();
logger.info("--> adding 2 nodes on same rack and do rerouting");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder()
.put(newNode("node1"))
.put(newNode("node2"))
).build();
RoutingAllocation.Result rerouteResult = allocation.reroute(clusterState);
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
// starting primaries
rerouteResult = allocation.applyStartedShards(clusterState, clusterState.routingNodes().shardsWithState(INITIALIZING));
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
// starting replicas
rerouteResult = allocation.applyStartedShards(clusterState, clusterState.routingNodes().shardsWithState(INITIALIZING));
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
logger.info("--> verifying all is allocated");
assertThat(clusterState.routingNodes().node("node1").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node1").get(0).state(), equalTo(STARTED));
assertThat(clusterState.routingNodes().node("node2").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node2").get(0).state(), equalTo(STARTED));
logger.info("--> adding additional node");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder(clusterState.nodes())
.put(newNode("node3"))
).build();
rerouteResult = allocation.reroute(clusterState);
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
assertThat(clusterState.routingNodes().node("node1").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node1").get(0).state(), equalTo(STARTED));
assertThat(clusterState.routingNodes().node("node2").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node2").get(0).state(), equalTo(STARTED));
assertThat(clusterState.routingNodes().node("node3").size(), equalTo(0));
String origPrimaryNodeId = clusterState.routingTable().index("test").shard(0).primaryShard().currentNodeId();
String origReplicaNodeId = clusterState.routingTable().index("test").shard(0).replicaShards().get(0).currentNodeId();
logger.info("--> moving primary shard to node3");
rerouteResult = allocation.reroute(clusterState, new AllocationCommands(
new MoveAllocationCommand(clusterState.routingTable().index("test").shard(0).primaryShard().shardId(), clusterState.routingTable().index("test").shard(0).primaryShard().currentNodeId(), "node3"))
);
assertThat(rerouteResult.changed(), equalTo(true));
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
assertThat(clusterState.routingNodes().node(origPrimaryNodeId).get(0).state(), equalTo(RELOCATING));
assertThat(clusterState.routingNodes().node("node3").get(0).state(), equalTo(INITIALIZING));
logger.info("--> fail primary shard recovering instance on node3 being initialized by killing node3");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder()
.put(newNode(origPrimaryNodeId))
.put(newNode(origReplicaNodeId))
).build();
rerouteResult = allocation.reroute(clusterState);
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
assertThat(clusterState.routingNodes().node(origPrimaryNodeId).get(0).state(), equalTo(STARTED));
assertThat(clusterState.routingNodes().node(origReplicaNodeId).get(0).state(), equalTo(STARTED));
}
@Test
public void deadNodeWhileRelocatingOnFromNode() {
AllocationService allocation = createAllocationService(settingsBuilder()
.put("cluster.routing.allocation.concurrent_recoveries", 10)
.put(ClusterRebalanceAllocationDecider.CLUSTER_ROUTING_ALLOCATION_ALLOW_REBALANCE, "always")
.build());
logger.info("--> building initial routing table");
MetaData metaData = MetaData.builder()
.put(IndexMetaData.builder("test").settings(settings(Version.CURRENT)).numberOfShards(1).numberOfReplicas(1))
.build();
RoutingTable routingTable = RoutingTable.builder()
.addAsNew(metaData.index("test"))
.build();
ClusterState clusterState = ClusterState.builder(org.elasticsearch.cluster.ClusterName.DEFAULT).metaData(metaData).routingTable(routingTable).build();
logger.info("--> adding 2 nodes on same rack and do rerouting");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder()
.put(newNode("node1"))
.put(newNode("node2"))
).build();
RoutingAllocation.Result rerouteResult = allocation.reroute(clusterState);
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
// starting primaries
rerouteResult = allocation.applyStartedShards(clusterState, clusterState.routingNodes().shardsWithState(INITIALIZING));
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
// starting replicas
rerouteResult = allocation.applyStartedShards(clusterState, clusterState.routingNodes().shardsWithState(INITIALIZING));
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
logger.info("--> verifying all is allocated");
assertThat(clusterState.routingNodes().node("node1").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node1").get(0).state(), equalTo(STARTED));
assertThat(clusterState.routingNodes().node("node2").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node2").get(0).state(), equalTo(STARTED));
logger.info("--> adding additional node");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder(clusterState.nodes())
.put(newNode("node3"))
).build();
rerouteResult = allocation.reroute(clusterState);
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
assertThat(clusterState.routingNodes().node("node1").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node1").get(0).state(), equalTo(STARTED));
assertThat(clusterState.routingNodes().node("node2").size(), equalTo(1));
assertThat(clusterState.routingNodes().node("node2").get(0).state(), equalTo(STARTED));
assertThat(clusterState.routingNodes().node("node3").size(), equalTo(0));
String origPrimaryNodeId = clusterState.routingTable().index("test").shard(0).primaryShard().currentNodeId();
String origReplicaNodeId = clusterState.routingTable().index("test").shard(0).replicaShards().get(0).currentNodeId();
logger.info("--> moving primary shard to node3");
rerouteResult = allocation.reroute(clusterState, new AllocationCommands(
new MoveAllocationCommand(clusterState.routingTable().index("test").shard(0).primaryShard().shardId(), clusterState.routingTable().index("test").shard(0).primaryShard().currentNodeId(), "node3"))
);
assertThat(rerouteResult.changed(), equalTo(true));
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
assertThat(clusterState.routingNodes().node(origPrimaryNodeId).get(0).state(), equalTo(RELOCATING));
assertThat(clusterState.routingNodes().node("node3").get(0).state(), equalTo(INITIALIZING));
logger.info("--> fail primary shard recovering instance on 'origPrimaryNodeId' being relocated");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder()
.put(newNode("node3"))
.put(newNode(origReplicaNodeId))
).build();
rerouteResult = allocation.reroute(clusterState);
clusterState = ClusterState.builder(clusterState).routingTable(rerouteResult.routingTable()).build();
assertThat(clusterState.routingNodes().node(origReplicaNodeId).get(0).state(), equalTo(STARTED));
assertThat(clusterState.routingNodes().node("node3").get(0).state(), equalTo(INITIALIZING));
}
}